Purpose: The wavelet transform is a newly developed signal-processing tool that decomposes a signal into various levels of resolution. The wavelet transform based noise reduction has the characteristics of optimally separating signal from noise, preserving the rapid rises and falls of a signal, and reconstructing a smooth signal from noise-imposed observations. The aim of this study was to evaluate the effects of applying a new noise reduction technique, the wavelet transform based noise reduction, to single photon emission computed tomography (SPECT) images.
Methods: Three experiments were performed using cylindrical phantom, line source, and hot-rod phantom, respectively. We acquired SPECT image datasets of each phantom, and reconstructed SPECT images using the wavelet transform based noise reduction with filter back projection (FBP). Images were de-noised by 3 parameters of wavelet transform based noise reduction: 1st wavelet weight (WW), 2nd WW, and 3rd WW, respectively. We evaluated the variances of full width at half maximum (FWHM), coefficients of variation (%CV), and frequency domains (radius direction distribution function in the power spectrum), respectively.
Results: In the cylindrical phantom test, %CV was reduced from 27.92% to 15.38% using the wavelet approach. On the other hand, FWHM values showed no significant change. However, the increases of wavelet weights caused artifacts on the reconstructed images in some cases.
Conclusions: The wavelet based noise reduction had the significant potential to improve SPECT image. Therefore, the wavelet method should prove to be a robust approach to improve image quantification and fidelity.